Peer review is not a guarantee of accuracy or truth. It's a relatively recent process that started from good premises, but in its current form has degenerated mostly into a conformity check. Peer review process might occasionally catch some egregious mistakes, or manipulations, but offers no incentives to actually test the results in a paper, and therefore verify its results. That would require access to experiment data (not all publications require authors to provide it, actually most don't) and then time to replay the experiment and check the result; as peer-reviewing is not paid, and counts for little in terms of career, very few reviewers bother.
The task of catching bad results falls therefore on the people who will attempt replication of the result after its publication, and sometimes it will take years before this happens.
Plenty of famous, even seminal peer-reviewed articles have subsequently been proved wrong, or even fraudulent. And, in general, a huge percentage of published results cannot be reproduced (General overview:
https://www.nature.com/articles/533452a.pdf . An egregious example from the field of cancer research:
https://www.nature.com/articles/483531a.pdf ).
So, the fact that a paper has been peer-reviewed means very little, it certifies little more than compliance with some very broad rules, and certainly says nothing about the truth of the content. The process is quite ineffective at its stated goal, and there is a growing movement of opinion to reform it, or even abolish it. After all, before WWII it did not exist, and plenty of good science was done nevertheless.IPB